Yuki Takahashi

Although the gender gap in mathematics and sciences in OECD countries is negligible, female students are still less likely to major in STEM fields in college. One explanation that has received less attention in the literature is that STEM programs are predominantly male-dominated, which makes female students a minority. We study whether the gender ratio at colleges affects high school students' college choices and the extent to which it contributes to the gender gap in STEM programs. We begin by using administrative data to show that the gender ratio is highly persistent over time, with almost no changes over 24 years. We then use an incentivized discrete choice experiment and show that the gender ratio at colleges does affect both female and male students' college choices: both female and male students prefer gender-balanced college programs over those with a male or female majority. Students avoid programs where they will be a minority mainly because they expect it to be difficult to fit into such environments. A counterfactual analysis suggests that their belief about the gender ratio in STEM programs reduces the likelihood of female students choosing STEM programs by 2.9 percentage points. Removing this constraint would increase the number of female students choosing STEM across the entire ability distribution and would improve utility by 0.26 standard deviations, as measured by program quality. Thus, the gender ratio at colleges is an important factor for high school students' college choices, and making STEM programs more gender-balanced can help narrow the STEM gender gap without sacrificing the quality of applicants.

In a laboratory experiment with eye tracking and using gender stereotypes, this paper demonstrates that decision-makers allocate less attention to negatively stereotyped individuals than to positively stereotyped individuals when finding the better-performing individuals and more attention to negatively stereotyped individuals than to positively stereotyped individuals when finding the worse-performing individuals. This selective attention has consequences for the accuracy of performance evaluations: decision-makers provide less accurate evaluations of negatively stereotyped individuals than of positively stereotyped individuals when finding the better-performing individuals. The differences in attention and evaluation accuracy are not driven by individuals’ actual performance. Showing continuous performance information does not override stereotypes. These findings suggest that stereotypes can lead to suboptimal selection of talent and harm negatively stereotyped individuals, such as when managers decide which workers to promote and dismiss.

While successful teamwork often involves correcting colleagues' mistakes, it may have negative interpersonal consequences. In an experiment, I show that it also has negative economic consequences: individuals are less willing to collaborate with those who have corrected them, even when the correction benefits the team. The data are consistent with negative feedback aversion: individuals who initially received positive feedback about their ability are significantly less willing to collaborate with those who corrected their mistakes, but not with those who corrected their right actions. Additionally, I find that men, but not women, are more tolerant of women who corrected their right actions. It is potentially due to men's beliefs about women's abilities, making women's corrections of their right actions less ego-threatening. This reluctance to work with those who provide corrective feedback can undermine teamwork, and mixed-gender teams may attract less competent women due to gendered sorting.

Light abuses and threats to receive them at home can deteriorate individuals' well-being, even in the absence of severe physical injury. Leveraging Russia's criminal law reform that decriminalized minor domestic violence, I first confirm that the number of domestic violence incidents classified as criminal offenses against female partners indeed decreased sharply after the reform. Using a difference-in-differences approach, I then show that the reform reduced married women's life satisfaction, increased depression, and increased college-educated married women's alcohol intake. Suggestive evidence indicates that the reform contributed to a decline in new marriages, while the divorce rate remained unchanged. These changes are unlikely to stem from shifts in violence outside the household, as there were no significant changes in gender-based violence or other crimes during the same period. These findings suggest that even minor intimate partner violence decreases married women's well-being and highlights the importance of legal institutions in addressing household violence.

Although evidence suggests men are more generous to women than to men, it may stem from paternalism and could reverse when women excel in important skills for one's career success, such as cognitive skills. Using a dictator game, this paper studies whether male dictators allocate less to female receivers than to male receivers when these receivers have higher IQs than dictators. By exogenously varying the receivers' IQ relative to the dictators', I do not find evidence consistent with this hypothesis; if anything, male dictators allocate slightly more to female receivers with higher IQs than to male receivers with equivalent IQs. The results hold both in mean and distribution and are robust to the so-called ``beauty premium.'' Also, female dictators' allocations are qualitatively similar to male dictators. These findings suggest that women who excel in cognitive skills may not receive less favorable treatment than equally intelligent men in the labor market.